How to Build a Serious-Incident Interview Plan in 24 Hours

OSHA tells employers to investigate fatalities, injuries, illnesses and close calls because incident investigation is where hazards and program gaps become visible before they repeat. This guide gives senior EHS leaders a 24-hour interview plan that protects witness memory, evidence quality and trust after a serious event.

Why does a serious-incident interview plan need to start before RCA?

A serious-incident interview plan needs to start before root-cause analysis because witness memory, site conditions and informal narratives begin changing within hours. OSHA's incident investigation guidance says investigations should look beyond immediate causes and avoid fault-finding, which makes the interview plan an evidence-control tool, not a blame session.

On the Headline Podcast, Tim Page-Bodoff framed the issue with a sharp line: investigation should chase a root-cause "what," not a root-cause "who." That distinction matters in interviews because the first version of the story often becomes the version everyone defends, especially when supervisors begin asking questions before the team has agreed on scope, sequence and language.

The practical move is to separate interviewing from judging. Treat the first 24 hours as a fact-preservation window that feeds the incident evidence map, not as the moment to decide who was wrong. The plan below gives EHS managers and senior leaders a sequence that can be used in manufacturing, energy, mining, logistics and construction.

Step 1: What decision must the interview plan protect?

The first decision is whether the interview process will protect learning or protect an early opinion. In the first hour after a serious event, the investigation lead should write a one-sentence purpose statement: preserve facts, understand decisions and identify conditions that made the event possible. That sentence becomes the boundary for every question asked in the first 24 hours.

Co-host Andreza Araujo's own work in A Ilusão da Conformidade argues that investigations should understand rather than punish, because the hunt for a culprit wastes the learning. Across the Headline Podcast conversations with Dr. Megan Tranter, the same leadership pattern appears: the quality of the first response determines whether people keep talking or start editing themselves.

Write the purpose statement before interviewing anyone, then share it with the investigation team and the site manager. A good version sounds like this: "This interview plan documents what each person saw, heard, decided and needed, so the organization can understand the event without assigning blame during fact collection." If the statement sounds like discipline preparation, rewrite it.

Step 2: Freeze the witness list before the story spreads

The witness list should be frozen within the first operational cycle because informal retelling contaminates sequence, timing and confidence. For a serious injury, process-safety near miss or high-potential event, the initial list should include direct witnesses, people who arrived immediately after, supervisors, permit issuers, maintenance planners, control-room staff and contractors whose work interface mattered.

What most investigations miss is the difference between a witness and a narrator. A witness has direct sensory contact with a fact. A narrator has a version of the event, often useful but more vulnerable to group memory. If the plan mixes those two groups, the team may mistake confidence for evidence and build a clean story that later conflicts with the chain of custody for incident evidence.

Create a two-column register with name, role, location, time observed, direct or indirect knowledge, language needs, fatigue status and any reporting-line conflict. In unionized or contractor-heavy sites, add representation needs before the interview begins. The register should be controlled by one person, since duplicate lists produce duplicate pressure.

Step 3: Separate interviews from medical care, discipline and notification

Interviewing must never compete with emergency response, medical care, family notification or legal reporting duties. OSHA's recordkeeping rule at 29 CFR 1904.39 sets specific reporting duties for fatalities, inpatient hospitalizations, amputations and eye losses, and those obligations should sit beside the interview plan rather than inside it.

This separation is cultural as much as procedural. In more than 250 cultural transformation projects, Andreza Araujo has observed that workers read leadership intent through the small order of actions. If the company interviews before caring for people, the stated purpose of learning will not be believed, even when the questions are technically neutral.

Set a rule that no injured worker is interviewed until medically cleared and emotionally able to participate. For direct witnesses who are shaken, begin with a short welfare check and schedule a later factual interview if needed. 24 hours is a planning window, not a license to pressure people, and the investigation leader must be able to defend that distinction.

Step 4: Build a neutral question set that reconstructs work as performed

A related Headline role profile on incident investigation facilitators in the first 30 days shows why question neutrality depends on the broader facilitation rhythm. Interviews work best when evidence, timeline, temporary controls and ownership decisions are kept separate.

A neutral question set reconstructs work as performed by asking what the person saw, expected, needed and chose at the time. The minimum set should cover sequence, task setup, controls, supervision, changes, signals, pressure, communication and stop-work options, with no question that assumes noncompliance before facts support it.

On a Headline Podcast episode, Dr. Thomas Krause noted that serious-event decision analysis often looks like employee fault until leaders examine conditions created months or years earlier. That is why the question "why did you fail to follow the procedure?" is inferior to "what made the procedure hard to follow at that moment?" The first closes the inquiry. The second opens the system.

Use a question bank, but do not read it like a script. Ask one question at a time, pause long enough for the person to think, and follow facts rather than your favorite theory. If a witness says, "we always do it that way," mark it as a system clue and compare it with training, permits, maintenance history and previous near-miss reports.

Step 5: How should the interviewer record uncertainty?

The interviewer should record uncertainty as evidence, not as weakness. Every interview note needs three labels: observed directly, heard from someone else, or inferred after the event. This simple marking method prevents confident speculation from entering the fact base as if it were confirmed sequence.

James Reason's Swiss cheese model remains useful here because serious events usually emerge through multiple failed layers, not one dramatic mistake. Co-host Andreza's book Sorte ou Capacidade makes the same point in practical language: an accident is the late result of barriers that failed, not a random strike. Interviews should therefore search for layer interaction rather than heroic certainty.

During note-taking, capture exact words when they concern time, alarms, instructions, permit conditions or perceived pressure. Avoid polishing grammar or replacing worker language with safety jargon. When the witness is unsure, write "unsure," then identify what evidence could confirm the point. This keeps the interview connected to the root-cause what, not who discipline.

Step 6: Protect contractor and supervisor interviews from power pressure

Contractor and supervisor interviews need extra design because economic and reporting pressure can distort what people say. The plan should identify who has authority over employment, contract payment, discipline and performance review, then keep those people out of the interview room unless a formal process requires their presence.

This is one place where senior leaders often underestimate the cultural signal. A contractor who depends on the client for the next purchase order may hear a neutral question as a threat. A supervisor may protect a production decision because they believe the organization rewards uninterrupted output more than honest escalation.

Use a two-person interview team with one safety lead and one operationally competent listener, but avoid including the direct manager of the interviewee. Offer language support, clarify that retaliation is unacceptable, and document any constraint that could affect candor. If the event involves mining, align the plan with MSHA expectations and the relevant site reporting duties, since MSHA's accident investigation procedures handbook shows how formal investigations preserve evidence, roles and sequence.

Step 7: Connect each interview to documents, controls and physical evidence

Each interview should end with an evidence connection, because memory alone cannot carry a serious investigation. The interviewer should ask which document, alarm, photo, permit, log, training record, maintenance ticket or physical condition could confirm or challenge the account, then assign ownership for retrieving that item.

OSHA's process-safety incident investigation language requires prompt investigation for events that resulted in, or could reasonably have resulted in, catastrophic release, and it includes findings, contributing factors and recommendations. In that environment, interviews are weak if they do not connect to controls. A statement about a valve position, isolation point or alarm response must be tied to evidence that can survive scrutiny.

Build an interview-to-evidence matrix with four columns: statement, evidence needed, owner and status. 48 hours is a useful upper limit for retrieving fragile evidence such as CCTV, digital logs, shift notes and contractor messages, because overwrite cycles and cleanup can erase context quickly. Link this matrix to your corrective-action tracker only after the fact pattern is stable.

Step 8: Close the 24-hour plan with a learning brief, not a verdict

The first 24 hours should close with a learning brief that states known facts, open questions, protected evidence and immediate risk controls. It should not name root causes, assign blame or present recommendations that the evidence has not earned. The brief is a bridge from preservation to analysis.

On Headline Podcast, Michael Emery has emphasized curiosity focused on what the line manager knows and what the operating department can actually change. That point is practical for the closeout brief: the people who run the work must recognize the findings as operationally real, or the corrective actions will become another safety document that nobody trusts.

Send the brief to the site leader, investigation sponsor, legal or compliance contact where required, and the operations owner of temporary controls. Include a short "do not conclude yet" note if leadership pressure is building. The plan succeeds when the next RCA session begins with cleaner facts, calmer witnesses and a clearer line between evidence and opinion.

Comparison: contaminated interviews vs. evidence-preserving interviews

Each day without a clear interview protocol makes the next serious event harder to learn from, because the organization will default to hierarchy, memory and pressure when it most needs disciplined curiosity.

Conclusion

A serious-incident interview plan is not an administrative courtesy after the event; it is the control that keeps early fear, hierarchy and hindsight from corrupting the fact base. The best 24-hour plan protects people first, then protects evidence, then gives the RCA team a cleaner starting point.

Headline Podcast is the space where leadership and safety come together to shape better workplaces and better lives. For more real conversations on how leaders respond when safety gets difficult, listen and subscribe at Headline Podcast.